-
-
Notifications
You must be signed in to change notification settings - Fork 401
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
scipy does not exit if there are build failures, but spkg-install looks OK #9519
Comments
This comment has been minimized.
This comment has been minimized.
comment:2
Could you give a link to the full log for scipy? |
comment:3
Well, Are you sure the finally produced(?) scipy is really broken? |
comment:4
Replying to @qed777:
Sure. Will attach to the ticket. Since its fairly large, I have compressed it. I think there is a problem with the build of ATLAS, as that shows a Dave |
Attachment: scipy-0.7.p5.log.gz Compressed version of sage-4.5/spkg/logs/scipy-0.7.p5.log |
comment:5
If enough parts of Sage run, you could try running the test suite, e.g., $ ./sage -sh
sage subshell$ easy_install nose
sage subshell$ exit
$ ./sage -python -c "import scipy; scipy.test()" Here's the output on sage.math with 4.5. I suppose we could optionally (if the user/developer desires) install nose automatically. |
comment:6
Replying to @nexttime:
I'm not certain. I thought ATLAS was broken, but I think I have remade the broken library. How can I test just scipy? The 64-bit SPARC port is not 100% ok. I can't run the doctests, as it segfaults, but Sage does semi-work. I can do computations with it. So Dave |
comment:7
Replying to @qed777:
I've never come across 'nose' - my python skills are next to zero. I will attach a log. As you can see, it finally fails with a core dump, but perhaps has some useful information before it dumps core. BTW, I've removed the static atlas libs, leaving only the shared ones. That might be a cause of a problem. I can't understand the need for both. If yyou look in the ATLAS package, you can see about 3 dynamic libraries will be missing. Two are not built, and one is deleted. I built them all and deleted no dynamic ones - only the static. Dave |
Attachment: scipy.test.log Output from: ./sage -python -c "import scipy; scipy.test()" > scipy.test.log 2>&1 |
comment:12
outdated, should be closed |
Reviewer: Dima Pasechnik |
comment:13
ok |
Building Sage 4.5 on a Sun Blade 2000, with dual UltraSPARC III+ processors in 64-bit mode, the build process produces some obvious error messages. These are not warnings, but errors.
wrong ELF class:
messages mean an attempt was made to link a mixture of 32-bit and 64-bit object files.But the build process still goes on to report that scipy has installed OK.
What is odd, is that
spkg-install
looks to be OK to me.The problem is not like the cephes package, or several others, where the return code of make is not checked.
At this point, I've not no idea if this is an upstream bug, or a Sage bug.
Anyone got any ideas?
Dave
Upstream: None of the above - read trac for reasoning.
CC: @nexttime @qed777 @sagetrac-mvngu @jhpalmieri @dimpase
Component: build
Reviewer: Dima Pasechnik
Issue created by migration from https://trac.sagemath.org/ticket/9519
The text was updated successfully, but these errors were encountered: